Re-engaging Justice-Involved Youth

130,000 juveniles are detained ever year, with nearly 70,000 juveniles in custody on any given day (Aizer and Doyle 2013). Prior justice involvement is one of the strongest correlates of adult criminal behavior (Sirakaya 2006). Given the magnitude of juvenile-justice involvement, it is imperative to understand what policies and programs can help justice-involved youth transition in to adulthood as productive, law-abiding citizens. In this dissertation, I will provide experimental evidence on the efficacy of two programs targeted at justice involved youth (14 to 22 years old) from two large scale randomized controlled trials conducted in Chicago in 2013 and 2014 by the City of Chicago and the Northern Illinois Project Safe Neighborhoods Task Force. Both programs have been shown to significantly reduce violence involvement among other populations (Heller 2013, Papachristos, Meares, and Fagan 2007); this dissertation will show if they can also reduce violence involvement among the group most likely to be involved in violence: justice-involved youth. The first of these programs, One Summer Plus, provides youth with 6 weeks of subsidized summer employment and social emotional programming. The second, Project Safe Neighborhoods (PSN) Youth Outreach Forums, requires youth probationers to attend a series of four forums designed to help youth recognize the legitimacy of the law and law enforcement and to internalize the consequences of crime. Half of males assigned to the forums will also be invited to participate in a 12 week cognitive behavioral therapy based program. The evaluations will use administrative arrest and schooling records to provide highly credible evidence about the programs' effects on arrests for violent crime, property crime, and other crimes, as well as school attendance and performance. We will measure the average effect of the program on the treatment group using an Intent-to-Treat setup. We will also identify the effect of treatment on youth who chose to participate, or the Local Average Treatment Effect, by using random assignment as an instrumental variable (lmbens and An grist 1994 ). ca/ncf